Perception of Innovative Usage of AI in Optimizing Customer Purchasing Experience within the Sustainable Fashion Industry

Abstract

The research was designed to contribute to scientific efforts in exploring the attitude of fashion stakeholders towards AI and its use in attaining sustainability in fashion industry. Although the role of AI in Fashion has been studied before, the aim of this research is to challenge and analyze the attitudes towards sustainable fashion of both stakeholders and consumers. The research considers the views of consumers, industry professionals and company shareholders on the role AI plays in pursuing ideas of Sustainable Fashion. Contrary to expectations, the companies with significant turnover did not show any greater awareness of the new trends in the fashion business. Furthermore, previous familiarity with the usage of AI did not prove to promote openness towards the recommendation of apps which use AI to promote Sustainable Fashion. The value of this research lies in the findings, which help provide a framework which can be used to change the viewpoint of the key market players. The crucial finding is that the AI approach on sustainability will influence both users (changing their purchasing decisions toward more sustainable choices if provided with a set of information on ecological impact, production choices), and corporate businesses (changing the overall business strategy, planning, marketing communication and production designs). The paper offers milestones for further research on synergies between AI, fashion industry lined with UNS SDGs and purchasing behavior.

1. Introduction

AI in Fashion Market Research estimates that worldwide spending on AI in Fashion will grow from USD229 million in 2019 to USD 1260 million by 2024, CAGR of 40.8 percent Report [1].

The internet has already played an instrumental role in connecting buyers and sellers across the globe [2]. The spread of the global pandemic, COVID-19, sped up the digitization process of the Fashion Industry. During the pandemic, fashion-related commercial activities such as visiting showrooms and attending fashion weeks became a predominantly online phenomenon [3].

A massive paradigm shift from being an experience-based business model to a highly data-driven business model is taking place in the fashion industry, with AI contributing to stepping up the entire process, having an impact in areas such as trend analysis, fashion recommendation, supply chain management, sales forecasting, and digitized shopping, etc. [4] Fashion companies in RTFS (Real Time Fashion Systems) will be able to use personalization and customization, and the focus will be on how to effectively match the design of clothing to the behavior and interests of users, according to consumers or prosumers [5]. AI has become part of how business is conducted in every industry, including fashion. When we look closely, we find that AI is used by almost every segment of the fashion value chain, from product discovery to robotic manufacturing [6].

With the default mode of shopping for consumers slowly shifting towards e-retail [7], there seems to be no better time to incorporate AI-based solutions towards achieving goals around sustainability in the fashion and retail industry.

This present study aims to explore the factors that could promote more sustainable behavior in the fashion industry. The motivation came from numerous research data that recognized the fashion industry as having a considerable share in environmental pollution. The goal was to examine the perception of the fashion industry (professionals, designers, and business owners) towards the role AI can play in promoting Sustainable Fashion.

2. The Numbers—Environmental Damage Caused by the Fashion Industry

According to the Geneva Environment Network [8]:

• Approximately 8–10% of all greenhouse emission gases come from the fashion industry. This figure is more than the combined emissions from all international flights and maritime shipping [9]. Without external intervention, this number will reach 26% by the end of 2050 [10].
• Every year, roughly 93 billion cubic meters of water are used to produce garments by the fashion industry [11].
• One truck full of clothes is burnt or dumped every second [12].
• Across the globe, the fashion industry contributes roughly 20% of industrial wastewater pollution [13]. 60% of the raw material made to produce garments comprises plastic [9].
• The weight of microfibers released into the water from washing clothes is 500,000 tons annually. Roughly, it amounts to disposing about 50 billion crushed plastic bottles in water [10].

In light of these facts, it is time for leaders, company owners, decision-makers, researchers, and transformative policymakers across the world not only to look closely into this problem but to work as interdisciplinary groups towards designing macro and micro solutions across the domain to start solving the challenges ahead.

The focus of this paper, as a contribution to the new concept of Industry 5.0 revolution where humans will be working alongside robots and smart machines, following 4.0, which conceptualizes rapid change in technology, industries, and societal patterns and processes [14,15], is to explore the usage and operational perception of AI as a tool to offer solutions for tackling problems of sustainability as it is applied to the fashion and retail industry. Another aim is to create a basis for future research on how business needs will align with UN Sustainable Development Goals and what the role of AI will be.

3. Introduction to AI

Simply speaking, AI, or Artificial Intelligence, is the ability of machines to mimic human thinking in various ways: decision making, communication, and reasoning, etc. [16]. Dauvergne [16] compares AI to a child whose thinking deepens with every bit of exposure he has had, every puzzle solved, every new word picked up. Except, in this case, AI doesn’t have any human needs or limitations of a human being and can learn indefinitely and capture infinite impressions, data points, and values of various kinds.

3.1. The Debatable Ethics of Use of AI

On the one hand, AI has been used for noble ventures such as preserving the coral reef off Australia’s Great Barrier Reef by sending Rangerbot, an underwater drone created for killing the crown-of-thorns starfish primarily responsible for the destruction of the coral reef to inject the starfish with poisonous bile salts, with an accuracy for identify the starfish of 99% and the risk of getting hurt by the poison on the starfish is negligible [16]. On the other hand, AI is also being used to create bots that persuade people of radical political agendas, an activity that was partially behind Trump’s presidential win [17].

Lahsen [17] elaborates in detail on the use of AI in current media structures used to shape public values, society, and a shared future for all. What that future would look like depends mainly on how well decisions and policymakers harness the use of AI. According to Kowalewski [18], “True, all humans do respond to the choices they have, but elites structure a framework of those choices.”

The onus on being transparent and providing sustainable options primarily lies with brands and manufacturers. According to Bolesnikov et al. [19], clients request a multi-level support system in various sectors, while this transition towards a ‘new symbiotic ecosystem’ takes place at the cost of traditionally established processes. It has become imperative for fashion brands to cater to the broadening scope of demands by conscious consumers [20].

3.2. Use of AI for Fashion Recommendations and Trend Forecasting

AI algorithms can be used to classify garment types and details, which can simplify and reduce the cost of trend forecasting [21].

In their paper, ‘Towards fashion recommendation: an AI system for clothing data retrieval and analysis,’ Kotouza et al. [22] propose the idea of creating an AI-based assistant for designers, manufacturers, and retailers to fulfill the demands placed on them due to the ever-evolving trends and tastes of clients (Fast Fashion). Such an AI tool (Figure 1) would need data clustering subsystems, web crawlers, sifting through metadata, etc., to develop design solutions that suit the client’s needs within the required time frame.

In their use case scenario, the AI tool (Assistant) uses Natural Language Processing [23] to transform the data sets it receives from two different sources into a specific format or design. The paper focuses heavily on demonstrating the benefits of using clustering techniques to enable AI to process an enormous amount of data at once and create options for designers to work with.

Similar data clustering techniques could be used for fashion recommendations, making the decision-making more accessible and more efficient for consumers, collaterally reducing wastage from erroneous or misinformed decision-making—uncontrolled and purchasing lacking planning.

In their paper, Kotouza et al. [22] prescribe this tool as a remedy for the demands made by Fast Fashion. A similar model could be built to develop AI to assist consumers in making sustainable choices (including slow fashion).

4. Literature Review

In their paper ‘fAshIon after a fashion—A Report of AI in Fashion’, Zou and Wong [24] coin the word ‘fAshIon’ to depict the use of AI in Fashion. They summarize the studies conducted over several decades into consolidated research by categorizing the use of AI in Fashion into seven groups: Overview, Evaluation, Basic Tech, Selling, Styling, Design, and Buying. The results aim to provide researchers with a holistic view of research in ‘fAshIon.’ The paper deals with the enormous potential that AI has created in fashion but fails to address sustainability as a topic. The authors believe that, considering the speed at which the industry is adapting to and investing in AI, this is the perfect time to promote ideas about using AI to promote sustainability in Fashion and Retail to the forefront.

How much consumers use AI is also going to predict whether they are inclined towards the use of AI Apps or not. Although consumers consider AI to be well equipped to handle low-complexity tasks involving problem-solving, consumers prefer humans over AI when it comes to customer service. The complexity of the task involved serves as a boundary in the perceived ability of AI to assist consumers [25].

4.1. The Pricing Dilemma

As a part of a comprehensive study carried out by Deniz Muslu [26], where she conducted multiple interviews with professionals from fashion brands, it came to light that companies encourage innovation when it brings efficiency to production processes, increases revenue, or conducts Corporate Social Responsibility (CSR) activities. Innovation around sustainability is not perceived favorably by firms when it suggests replacing existing systems with newer models.

In the same study, companies brought up the issue of the availability of sustainable materials, which are in short supply and tend to cost more. For the cost to come down, suppliers would require more demand from companies that look towards consumers to increase the demand. Since consumers expect retailers to offer sustainable products at reasonable or fair prices and to make sustainable purchase decisions, this makes for a chicken-and-egg problem, which continues to be a vicious cycle.

4.2. Impact of Consumer Awareness on Consumer Purchase Behaviour

A qualitative study carried out by Birstwistle and Moore [27] identified a lack of awareness among consumers regarding how their purchase behavior impacts the environment. Fortunately, we have come a long way from there.

Although in their study Hur and Cassidy [28] shortlisted a few internal and external challenges that the consumers and designers perceive while incorporating sustainable fashion choices, the study concluded that most participants felt that if they were provided with information about the ecological risks, they would change their purchasing behavior to avoid buying garments whose production harms the environment more and buy garments whose production is less damaging to the environment.

Awareness is all about relationships, as business results can be seen as a consequence of the above. For example, the tourism destination, modelled as a complex network system, is an excellent basis for developing numerical simulations to measure the potential of the established relationships. This methodology is increasingly imposing itself as a tool of support for the analysis and planning of complex social and economic systems, for which it is usually harder to apply techniques that are more traditional [29,30]. The same can be applied when researching purchasing behavior and decision to spend on fashion garments.

5. Goals and Hypotheses of Our Study

With all this in mind, our study aimed to examine the determinants of environmental awareness, i.e., willingness to use technologies to increase sustainability and employees’ attitudes in the fashion industry regarding AI transactions. Therefore, we explored the factors that could promote more sustainable behavior in the fashion industry. Being aware of numerous research data that show the fashion industry to be a major contributor to environmental pollution, we examined the perception and attitudes of the fashion industry (professionals, designers, and business owners) towards the role AI can play in promoting Sustainable Fashion. In addition, we explored the role of providing information about the environmental impact of garments production in promoting more environmentally conscious purchase behavior.

From our goals and previous literature review, we composed several hypotheses:

H1: 

Employees employed by companies with revenue of more than 10 million euros are more aware of global sustainability efforts.

H2: 

Those already using AI are more open to the innovative use of AI Apps that suggest fashion recommendations.

H3: 

The participants believe that use of AI in fashion can optimize production of fashion garments perception.

H4: 

Those aware of new trends in the fashion and retail industry are more open to the use of new technologies.

H5: 

The participants believe that use of AI and intelligent apps can optimize purchasing behavior.

H6: 

The ecological impact of production determines the purchasing decisions of fashion garment buyers.

6. Materials and Methods

6.1. Sample

In this research, a total number of 263 respondents filled in the questionnaire.

Table 1. Frequency Distribution of Companies by the Country of Origin.

Country/Continent/Region	Frequency	Percent	Valid Percent
India	137	52.1	58.8
EU	30	11.4	12.9
USA	17	6.5	7.3
Africa	9	3.4	3.9
Switzerland	7	2.7	3.0
SE Asia countries	6	2.3	2.6
Balkan countries (non-EU)	6	1.9	2.1
China	5	2.3	2.6
UK	4	1.5	1.7
ME countries	4	1.5	1.7
Canada	2	0.8	0.9
UAE	2	0.8	0.9
Russia	2	0.8	0.9
Former Soviet Union countries	2	0.8	0.9
■ Total	233	88.6	100.0
■ Missing	30	11.4
■ Total	263	100.0

shows the country of origin (30% of respondents did not answer this question). More than half of the participants represented companies from India (58.8%), 12.9% represented EU companies, 7.3% companies from the USA, 3.9% from Africa, and 3% from Switzerland. Other countries made up less than 3%.

The distribution of the participants by job role is presented in Figure 2. Most of the respondents held managerial positions (38%), 15.2% were executives, 16.7% were owners, and 30% had other job roles. Figure 3 depicts the distribution of the companies by the number of employees. Almost 45% of participants work in companies with more than 100 employees, 19.8% had between 50 and 100 staff members, 17.3% had less than 50, and 18.3% had less than 10 employees.

The companies’ share based on their revenue is shown in Figure 4. Almost half of the respondents work at companies whose revenue was lower than 500 thousand euros (47.5%), 17.5% of companies had a revenue of over 100 million euros, 10% had a revenue between 1 and 10 million euros, 10.3% had a revenue between 500 hundred and one million euros, 9.5% had revenue between 10 and 50 million euros, and 4.2% had a revenue between 50 and 100 million euros. In Figure 5, it can be seen that 82% of the companies make their profit from services.

6.2. Instruments

As a research instrument consisting of a series of questions to gather information from respondents, this questionnaire was created and sent out to an extensive network of authors’ contacts via academic institutions they work with and via the LinkedIn business social network. The questionnaire was made using Google Docs.

Through the questionnaire we explored the opinions of the owners and employees in the fashion industry towards using AI and downcycling. The questions were designed to evaluate the perception and openness of the fashion industry’s professionals towards AI and its use in promoting Sustainability in Fashion. This questionnaire is part of the Appendix A.

6.3. Variables

With the exception of the income estimate and ratings of awareness of new trends, which were ordinary variables, all other variables were nominal (yes/no/maybe types of answers).

6.4. Statistical Analysis

Since the research was mainly explorative, the frequencies and percentages of the answers were calculated, and accordingly, in all analyses we conducted Chi-square tests. Additionally, for testing H1, the independent-samples t-test was calculated. All data analyses were run in SPSS for Windows.

7. Results

To test H1, the owners were excluded from the sample since we wanted to test only the employees’ perceptions. The companies were divided into two groups, the first group consisted of those whose revenue was less than 10M (N = 139), and the second group contained those whose revenue was greater than 10M (N = 80). The dependent variable was the self-reported awareness of the importance of new trends in the fashion and retail industry. Contrary to our hypothesis, companies with a revenue of less than 10 M had a greater level of awareness of the importance of new trends (M = 3.37, SD = 1.11) compared to companies with a revenue greater than 10M (M = 3.11, SD =1.09). These group differences in the level of awareness are shown in Figure 6. However, the t-test did not show statistically significant differences: t (217) = 1.64, p = 0.102. Thus, we rejected the first hypothesis that the companies with the larger revenue will show greater awareness of the new trends in the fashion industry.

With testing H2, we wanted to explore the differences in the openness to the use of fashion garments between participants based on the experience with AI. Figure 7 and Figure 8 show the differences between participants who already use AI in their work ventures and those who do not. Figure 6 shows that those who use AI in their work are more open to using an app to optimize their wardrobe stock. However, the chi-square is not statistically significant: χ² (1) = 1.5, p = 0.22. Figure 8 shows the differences in openness to receiving AI suggestions for future purchases between the participants who use AI and those who do not use it. Based on the answers, it is notable that those experienced with AI are more open to apps based on AI. However, when tested with the chi-square test, this difference is not significant: χ² (1) = 2.78, p = 0.10. Based on these results, we rejected H2 since two tested groups of participants are statistically equally open to new AI apps.

With the H3, we explored the differences between respondents who already use AI and those who do not, believing that using the apps based on AI can improve participants’ everyday business operations and purchase behavior. Figure 9 shows the distribution of frequencies of the answers related to the participants’ belief that AI can help them in day-to-day business operations and participants’ usage of AI. Although both groups of participants believe that AI can improve their work experience, participants whose companies already use AI are more prominent in the beliefs of AI benefits. The difference in frequency is statistically significant: χ² (2) = 84.36, p < 0.001.

The differences in answers between AI and non-AI users concerning the beliefs that AI could improve their purchasing behavior are presented in Figure 10. We can see that 86.7% of AI users believe AI is beneficial for purchasing behavior, although 62.2% do not use AI in their companies. Furthermore, non-AI users tend to be more uncertain about AI benefits (30.1% responded that AI might improve their purchases). The differences in the distribution of the frequencies are statistically significant: χ² (2) = 20.3, p < 0.001. The results we recorded enable us to accept H3; thus, we were able to conclude that those already using AI in their work are more likely to believe that AI can optimize the production of fashion garments by optimizing their business operation and purchasing behavior.

To test H4, we cross-tabulated the awareness ratings of the importance of new trends in the fashion industry with the answers on several items. The results are shown in Figure 11 and Figure 12. We recorded statistically significant differences in the distribution of the answers in two of the four analyses. There were no significantly different distributions of the ratings of the awareness of new technologies considering respondents’ belief that AI smart app could improve their purchasing behavior (χ² (8) = 8.01, p = 0.43) and the item exploring the openness towards apps that optimize their wardrobe stock (χ² (4) = 9.10, p = 0.06). However, those participants who rated their understanding of new trends with higher grades are more familiar with the business model of downcycling: χ² (4) = 40.0, p < 0.001 (Figure 11). Similarly, more participants are open to AI suggestions based on their wardrobe stock and previous purchases among those who rated their awareness of new trends with higher grades: χ² (4) = 11.42, p < 0.05. Based on these results, we can partially accept H4 that those aware of new trends in the fashion industry are more open to the use of new technologies.

In H5, we supposed that AI apps could optimize purchasing behavior, i.e., we wanted to explore participants’ opinions toward this claim. In

Table 2. The answers to the items exploring participants’ belief in AI apps and openness towards their usage for optimization of their wardrobe stock.

Items	Yes		No		Maybe		χ²
	f	%	f	%	f	%
Do you believe AI and smart apps could improve your purchasing behavior?	193	73.38	13	4.94	57	21.67	200.88 *
Would you be open to using an app optimizing your wardrobe stock?	192	73.00	71	27.00	0	0	55.67 *
Would you be open to receiving AI suggestions on your future purchases based on your wardrobe stock, your recorded purchase, and current online offers?	200	76.05	63	23.95	0	0	71.37 *

Notes: f—frequency; %—per cent; *—all tests are significant at p < 0.001.

, we presented the answers to three items. The first item states that participants believe that AI smart apps can improve their purchasing behavior; the second one explores their openness towards the usage of AI apps for the optimization of their wardrobe stock; finally, with the third item, we tended to find out whether participants are open to receiving AI suggestions on their future purchases. The distribution of frequencies from Table 2 advances our predictions; thus, most respondents agree with all three items, and the differences tested with the Chi-square test are statistically significant (Table 2). Based on the results recorded, we can accept the fifth hypothesis. Therefore, we can conclude that participants are confident that the usage of AI can optimize their purchasing behavior and wardrobe stock.

Finally, in the H6, we tested whether the awareness of the ecological impact of the wardrobe industry shapes participants’ purchasing decision of fashion garment. In

Table 3. The distribution of answers on the items exploring participants’ purchasing tendencies based on the ecological impact of the clothes production.

Items	Yes		No		χ²
	f	%	f	%
Would you be buying less “damaging” clothes if you were aware of the impact?	235	89.35	28	10.65	162.92 *
Would a standardized declaration measuring the damage caused by every garment piece manufactured change your purchasing behavior?	222	84.41	41	15.59	124.57 *
Do you believe natural materials cause minor environmental damage if used in fashion garments production?	239	90.87	24	9.13	175.76 *

Notes: f—frequency; %—per cent; *—all tests are significant at p < 0.001.

, we presented the frequency of the answers on three items. Most participants agree with all statements. In other words, if they are provided with information about the ecological risks, they would change their purchasing decisions. This result is significant since it implies that informing people about the environmental impact of their clothes would probably change their purchasing behavior (i.e., they would probably avoid buying clothes whose production damages the environment). Based on Chi-square tests, the difference in frequency is statistically significant. Therefore, we can accept H6 and conclude that the ecological impact of production could determine buyers’ purchasing decisions if they are provided with such information.

8. Discussion

Our research tended to investigate the factors that could enhance more sustainable behavior in the fashion industry. We mainly focused on employers’ and employees’ perceptions and opinions towards the usage of AI in their everyday business and their purchasing behavior of fashion garments. Finally, we explored their awareness of the importance of downcycling and what could trigger a change in buying behavior in terms of persuading them to buy eco-friendly products.

Contrary to our expectations, the companies with the more significant income did not show greater awareness of the new trends in the fashion industry. This shows the necessity for policymakers to impose economic sanctions on those opposing becoming a sustainable organization. Furthermore, we did not confirm the assumption that previous familiarity with the use of AI will be a factor in promoting openness towards recommendations of AI Apps. Obviously, businesses are eager to use AI and new technologies in general to boost their revenues, but not to introspectively redesign themselves into more sustainable, environmentally aware and socially responsible organizations.

It is almost disturbing to see the gap businesses have in their mindsets when it comes to the creation of additional value for themselves and their shareholders versus the creation of new value for the global, eco-aware movement. However, this is the case with other industries as well. A good example is the research carried out in the tourist industry in Italy. In the case study analysis, it emerged that, within family businesses in the tourist industry, innovation is limited and is not the only driver to ensure sustainable development [31]. Although UN goals deal with long-term planning, it is noticeable that companies, business owners and stakeholders in fashion are short-sighted when not wanting to change. Having a scenario where a legal entity or a brand decides not to invest now to obtain security in production and sales in the future versus having unprofessional hopes “something” or “someone” will sort the issue out for them shows how much can be done in the field of sustainable education and the introduction of sustainable culture in the fashion industry.

However, employees that use AI in their work are more disposed to believing that AI can optimize the production of fashion garments by optimizing their business operation and purchasing behavior. Additionally, the awareness of new trends can assist the usage of new technologies. Participants, in general, believe that the use of AI can optimize their purchasing behavior and help manage their existing wardrobe stock. The most important part of this study is the finding that participants are willing to change their purchasing decisions to more sustainable choices if provided with information about the ecological risks. These results are in accordance with previous research by Hurr and Kassidy [28]. Further, as Diaz et al. found [31], to speed up and promote sustainable behavior, policymakers should provide enough information through education, and they should also motivate citizens by making them aware of how they can contribute to the environmental cause.

9. Limitations

The study has limitations in line with the vast field of research of AI, fashion, sustainability, and business. Any business, being a live organism, changes, evolves or dies in the circle of life. The same goes, but at a faster pace, for the application of innovative models of communication, relationships, and goals.

The study does not measure the beliefs of the general population and only focuses on measuring the belief and the perceptions of the professionals in the fashion industry. It merely discusses potential outcomes on the decisions to purchase in retail based on sustainable goals.

The data gathering sources were online portals and hence were limited to users of those online portals or social media websites.

While conducting the research, authors realized that income per user or income per purchase would have been a better choice than merely presenting revenue as a symbol of a size of a business (company). Future work will focus more on the exploration of the quality of sales (margins and size of the revenue per sale) in relation to the use of AI and digital customer experience (online sales) to understand the impact of sustainability communication. Authors are hoping to continue the research by finding out more about spending versus quality versus origin of materials versus sustainability relations.

10. Conclusions

Our study aimed to examine the determinants of environmental awareness, i.e., willingness to use technologies to increase sustainability and employees’ attitudes in the fashion industry regarding AI transactions. Although our research did not consider the general population’s opinion, we were able to conclude that sustainability has become an essential factor impacting purchasing behavior in the fashion industry.

There is an excellent chance to get to the point where AI in fashion can direct the brands and consumers towards making sustainable choices. The process will bring end-users to hassle-free experience, more optimized garment stock, more household space, a chance to buy more quality while buying less, collaterally leading to less pollution while working with the intent to increase the expected life frame of the garment.

As innovative apps technology is becoming more approachable and affordable, interested parties can use testing without jeopardizing cash flows and budgets. Finally, the concept of marketing messaging will change as we see it, from “come and buy,” to “let’s make conscious choices together”.

Further research is required to study the specifications that would help define the kind of AI Apps that satisfy consumer expectations such as reducing the inconvenience of data entry involved in using AI Apps to manage a wardrobe. Studies are also required to introduce more sustainable and natural production materials with the business goal of profitability and the API possibilities of merging the activities and data of different AIs to increase accuracy and effectiveness.